Академический Документы
Профессиональный Документы
Культура Документы
OMAE2008
July 15-20, 2008, Estoril, Portugal
OMAE2008-57893
THE INFLUENCE OF FOUNDATION MODELING ASSUMPTIONS ON
LONG-TERM LOAD PREDICTION FOR OFFSHORE WIND TURBINES
Erica Bush
University of Texas at Austin
Austin, TX, USA
Puneet Agarwal
University of Texas at Austin
Austin, TX, USA
ABSTRACT
In evaluating ultimate limit states for design, time-domain
aeroelastic response simulations are typically carried out to
establish extreme loads on offshore wind turbines. Accurate
load prediction depends on proper modeling of the wind
turbulence and the wave stochastic processes as well as of the
turbine, the support structure, and the foundation. One method
for modeling the support structure is to rigidly connect it to the
seabed; such a foundation model is appropriate only when the
sea floor is firm (as is the case for rock). To obtain realistic
turbine response dynamics for softer soils, it is important that a
flexible foundation is modeled. While a single discrete spring
for coupled lateral/rotational motion or several distributed
springs along the length of the monopile may be employed, a
tractable alternative is to employ a fictitious fixed-based pile
modeled as an equivalent cantilever beam, where the length
of this fictitious pile is determined using conventional pile
lateral load analysis in combination with knowledge of the soil
profile.
The objective of this study is to investigate the influence of
modeling flexible pile foundations on offshore wind turbine
loads such as the fore-aft tower bending moment at the
mudline. We employ a utility-scale 5MW offshore wind
turbine model with a 90-meter hub height in simulations; the
turbine is assumed to be sited in 20 meters of water. For a
critical wind-wave combination known to control long-term
design loads, we study time histories, power spectra, response
statistics, and probability distributions of extreme loads for
fixed-base and flexible foundation models with the intention of
assessing the importance of foundation model selection. Load
distributions are found to be sensitive to foundation modeling
assumptions. Extrapolation to rare return periods may be
expected to lead to differences in derived nominal loads needed
in ultimate limit state design; this justifies the use of flexible
foundation models in simulation studies.
Lance Manuel
University of Texas at Austin
Austin, TX, USA
INTRODUCTION
Nominal loads for the design of wind turbines in ultimate
limit states are generally established from time-domain
aeroelastic response simulations. The accuracy of these derived
loads depends on the number of simulations and on how
realistically the models used to represent the turbine, support
structure, and foundation describe the true structural response.
One potential shortcoming in modeling foundations relates to
their flexibility. A single pile (often referred to as a monopile)
is the most common type of foundation used today for offshore
wind turbines; the support structure connects to such a pile
foundation that extends some depth below the mudline. One
way a monopile foundation could be modeled is by means of a
rigid connection at the mudline. This model ignores the soil
profile and the associated soil-pile stiffness and, as such, would
not account for the piles expected lateral/rocking movement.
Such simplifying assumptions could only adequately simulate
the behavior of a monopile founded in rock. Many offshore
wind turbines, however, are founded on softer soils where the
monopile experiences at least some movement at and below the
mudline. It is therefore worth assessing the accuracy of the use
of a fixed-base model versus a flexible foundation model. In
the present study, we carry out fixed-based model simulations
and study turbine loads (specifically, the fore-aft tower bending
moment). These are compared with loads derived using a
flexible foundation model. This latter model utilizes stiffness
properties derived from the soil profile at the location of the
turbine by means of a conventional pile foundation analysis and
appropriate p-y lateral load-deflection relationships. The
flexible foundation model involves derivation of an apparent
fixity length representing a distance below the mudline where
an equivalent cantilever yields the same lateral movement and
rotation as the monopile experiences in the pile analysis with
the true soil properties. The mass per unit length of the
equivalent cantilever is adjusted to match the sub-soil mass of
Turbine
tower
Mean
sea level
Mudline
Monopile
support
Fixed
connection
at the base
Apparent
fixity length
fixity length (l) and is derived along with the flexural rigidity
(EI) of the cantilever using the following equations:
Fl 3 Ml 2
w=
+
3EI 2 EI
(1)
Fl 2 Ml
=
+
2 EI EI
where w is the deflection and is the rotation at the mudline.
To model this new fictitious pile in FAST, its length, flexural
rigidity, and mass distribution are required. The length and
flexural rigidity are determined as described by Eq. (1). The
mass distribution is kept the same as the mass per unit length of
the monopile above the mudline. The mass of the equivalent
cantilever also closely matches the sub-soil mass of the original
pile and, thus, realistically accounts for inertia effects in the
flexible foundation model.
To arrive at values of F and M for the pile analysis and for
computation of the apparent fixity length, l, 150 ten-minute
simulations were run for the fixed-base case. Time histories as
well as summary statistics for F and M were obtained;
ensemble averages over the 150 simulations of the mean (),
standard deviation (), and maximum (max) values of the
shear, moment, and axial force at the mudline were computed.
Table 1 summarizes derived apparent fixity lengths and flexural
rigidity values for flexible foundation models that were
developed using three different fixed-base mudline force
combinations: (a) using mean () values; (b) using mean plus
one standard deviation values ( + ); and (c) using maximum
values (max). The deflections and rotations at the pile head
shown in Table 1 were based on LPILE analyses. As can be
seen from Table 1, the apparent fixity lengths and flexural
rigidities for the cantilever that serves as a representation of the
flexible foundation are only slightly different for the three
mudline force cases. This suggests that the soil behavior is
almost linear for the pile and soil properties (i.e., the profile)
studied here and for the range of forces encountered in
simulations with the selected wave height and wind speed. The
apparent fixity length varies between 17 and 18 meters.
Indeed, if the shear, moment, and axial force values are allowed
to take on a wide range of values that were obtained in the
fixed-base simulations, with representative contemporaneous
values of the forces taken randomly at different times during
each ten-minute simulation, the apparent fixity lengths
remained in a narrow range. This can be confirmed by studying
Fig. 3 which summarizes computations of the apparent fixity
length, l, for 50 randomly drawn contemporaneous shear,
moment, and axial forces from simulations with Hs = 5.5m and
V = 16 m/s. Clearly, greater than 60% of the time, the apparent
fixity length is between 17 and 18 meters. Also, in the range of
most likely shear force values, the apparent fixity length is
close to 17 meters; only for small shear values does this
apparent fixity length reduce by more than a meter. We might
note here that only for larger wave heights and/or larger wind
speeds, shear forces might be expected to be larger and, thus,
result in greater apparent fixity lengths in flexible foundation
models.
Table 1: Summary of the derived flexible-foundation models based on mudline forces at three different levelsmean (), mean + 1
standard deviation ( + ), and maximum (max). The mudline forces used in the three cases are M (moment) and F (shear); the
mudline lateral displacement (w) and rotation () at the top of the file are computed using LPILE. The derived apparent fixity length
(AFL) and flexural rigidity of the pile (EI) are also shown.
Force
basis
Case
1
2
3
+
max
AFL
EI
(MN-m)
(MN)
(m)
(rad)
(m)
(MN-m2)
45.15
57.08
90.59
0.40
1.02
2.45
0.0065
0.0092
0.0167
-0.0007
-0.0010
-0.0017
17.20
17.50
17.94
1,125,200
1,142,800
1,155,400
60
80
100
40
20
0
12
13
14
15
16
17
18
l (m)
(a)
18
16
Flexible - Mean
Flexible - Mean+
Flexible - Max
Rigid Foundation
10
10
-2
10
-4
10
0.5
1.5
2.5
Frequency (Hz)
14
12
-1.0
TBM
10
0.0
1.0
2.0
(b)
Figure 3: Study of apparent fixity lengths in flexible foundation
model based on random samples of contemporaneous shear
force, bending moment, and axial force at the mudline for V = 16
m/s and Hs = 5.5 m) (a) likelihood of different derived apparent
fixity lengths; (b) variation of apparent fixity length with mudline
shear force.
NUMERICAL STUDIES
A total of 150 ten-minute simulations were run for the
fixed-based model and for the flexible foundation model for the
selected environmental condition corresponding to a hub-height
ten-minute average wind speed, V, of 16 m/s and a significant
wave height, Hs, of 5.5 m. Our interest is primarily in the foreaft tower bending moment at the mudline. Figure 5 shows
representative 200-second segments from a single ten-minute
simulation; longitudinal wind speed, sea surface elevation, and
the fore-aft tower bending moment for the fixed-base and
flexible foundation models are shown. The wave input has an
influence on the tower loads as is evident from the figure.
Upon comparing fixed-base and flexible foundation tower
loads, it can be seen that the energy at higher frequencies is
diminished and the energy at some intermediate frequencies is
enhanced with the flexible foundation model. This can be
confirmed by also studying the power spectra in Fig. 4. The
rigid fixed-base model clearly has greater high-frequency
Figure 5: A representative 200-second segment of the hubheight longitudinal wind speed, the sea surface elevation, and
the fore-aft tower bending moment for two cases (fixed-base
and flexible foundation models) taken from a single 10-minute
simulation for V = 16 m/s and Hs = 5.5 m. The flexible
foundation models derived stiffness is based on mudline forces
derived using mean + 1 standard deviation forces from the
fixed-base case.
1
Fixed Base
Flexible Foundation
0.1
0.01
0.001
75
85
95
105
115
125
Table 2: Ensemble averages of various statistics of the fore-aft tower bending moment based on 150 ten-minute simulations. Peak
factors (PF) are computed by subtracting mean values from maxima/minima and dividing by the standard deviation.
Foundation
Model
Max
Min
Mean
Std. dev
(MN-m)
(MN-m)
(MN-m)
(MN-m)
Fixed-Base
90.6
6.4
45.2
Flexible
92.3
3.2
45.5
Skewness
Kurtosis
PF(max)
PF(min)
11.9
0.22
3.25
3.81
-3.25
13.0
0.09
3.08
3.60
-3.25
CONCLUSIONS
A flexible foundation model has been employed to study
extreme tower loads for an offshore wind turbine. The study is
motivated by the need to assess the degree of influence on loads
of foundation modeling assumptions and to assess the accuracy
of simpler fixed-base models for foundations that do not
account for the soil-pile stiffness in typical monopile
foundations. The flexible foundation model makes use of
details related to the soil profile, p-y curves for lateral response
of the soil-pile system, and the notion of an apparent fixity
length or distance below the true mudline that is derived on the
basis of the nonlinear p-y curves and realistic applied forces at
the mudline.
Time-domain simulations were carried out and show that
there is reduced high-frequency energy in turbine loads when
foundation flexibility is taken into account. Load extremes are
seen to be slightly larger for flexible foundations; the standard
deviation (variability in the load process) is also somewhat
larger and to a greater degree than for extremes. Probability
distribution curves for tower bending moment estimated from
simulations show greater variability for flexible foundations;
extrapolation to rare return periods is expected to lead to even
wider differences between fixed-base and flexible foundation
model predictions.
ACKNOWLEDGMENTS
The authors would like to acknowledge the National
Science Foundation for financial support by way of CAREER
Award No. CMMI-0449128 and Award No. CMMI-0727989.
We would also like to acknowledge Dr. Jason Jonkman at the
National Renewable Energy Laboratory for his continued
assistance with the program, FAST, and the wind turbine
simulation model used in this study. Finally, we would like to
thank Ensoft, Inc. for providing LPILE for our use and Dr. Shin
Tower Wang for his assistance with that program.
REFERENCES
[1] Agarwal, P. and Manuel, L., 2007, Simulation of Offshore
Wind Turbine Response for Extreme Limit States,